Appropriate noise addition to metaheuristic algorithms can enhance their performance

Area of Science:

• Computational intelligence
• Optimization algorithms
• Nature-inspired computing

Background:

• Swarm-based algorithms are powerful for complex optimization but suffer from premature convergence.
• Ensuring convergence to a global optimum remains a challenge for these general-purpose methods.

Purpose of the Study:

• To propose a novel strategy, heterogeneous Perturbation-Projection (HPP), to enhance the exploration capability of swarm algorithms.
• To guarantee sufficient convergence conditions and ensure algorithms converge to a global optimum.

Main Methods:

• The HPP strategy involves applying stochastic perturbation to half of the swarm agents.
• All agents are then projected onto the set of feasible solutions.
• The HPP strategy was tested with Particle Swarm Optimization (PSO), Bat Algorithm (BAT), and Ant Colony Optimization (ACO).

Main Results:

• Algorithms enhanced with HPP demonstrated significantly improved performance compared to their original versions.
• The HPP-enhanced algorithms outperformed the originals in 60-80% of experimental runs.
• Significant performance margins were observed in extensive numerical experiments.

Conclusions:

• The heterogeneous Perturbation-Projection (HPP) strategy is a general, simple, and effective method for improving swarm-based optimization algorithms.
• HPP guarantees convergence to a global optimum and overcomes common drawbacks like premature convergence.
• This approach offers a robust enhancement for various nature-inspired swarm intelligence algorithms.